JP6460099B2 - Packet transmission apparatus, packet transmission system, and packet transmission method - Google Patents

Description

  The present invention relates to a packet transmission device, a packet transmission system, and a packet transmission method, and more particularly to a packet transmission device, a packet transmission system, and a packet transmission method that can transmit packets with low delay and wide bandwidth.

  An example of a conventional packet transmission system is described in Non-Patent Document 1. The conventional packet transmission system shown in FIG. 15 includes a network node 7-A, a network node 7-B, a network node 7-C, and a network 3 connecting them. Here, for convenience of explanation, the case where there are three network nodes is shown, but the number is not limited. The network node 7-A, the network node 7-B, and the network node 7-C communicate with each other via the network 3.

Next, the configuration of the network node 7-A will be described with reference to FIG. The configuration of the other network node 7-B and the configuration of the network node 7-C are the same as the configuration of the network node 7-A. The network node 7-A includes a packet output application 71, a packet aggregation queue 72, a packet aggregation unit 73, a memory 74, and a packet reception unit 75. The packet output application 71 outputs a packet. The packet aggregation queue 72 stores packets to be transmitted to the network 3. The packet aggregating unit 73 aggregates a plurality of packets stored in the packet aggregation queue 72 and transmits them to the network 3. The memory 74 holds operation parameters of the packet aggregation unit 73. The packet receiving unit 75 receives a packet from the network 3 and passes it to the packet output application 71.
The packet aggregating unit 73 includes a timer 731 that detects a timeout that gives an opportunity to transmit the aggregated packet. The memory 74 holds a minimum aggregation number 741 indicating the minimum value of the aggregate packet number and a timeout value 742 indicating a packet aggregation timeout value. The packet 721 is the head packet of the packet aggregation queue 72, and the packet 722 is the same packet as the destination of the packet 721.

  The operation of the conventional packet transmission system having such a configuration will be described with reference to FIG. Here, an operation in which the packet output application 71 held by the network node 7-A transmits a packet to the network 3 will be described.

  The packet output by the packet output application 71 is stored in the packet aggregation queue 72. The packet aggregating unit 73 refers to the packet aggregating queue 72, a packet 721 exists at the head of the queue (denoted as packet P in FIG. 16), and the packet 722 has the same destination as the packet 721 stored in the packet aggregating queue 72 If the number (denoted as packet P ′ in FIG. 16) is larger than the minimum aggregation number 741 (step F101), a packet in which the packets 721 and 722 are aggregated is created (step F102). Here, the number of packets 722 (packets P ′) to be aggregated into the aggregated packet is the maximum value of all the packets 722 or the aggregation number allowed by the maximum packet length.

Next, the packet aggregating unit 73 refers to the packet aggregation queue 72, and when the packet 721 exists at the head of the queue and the timer 731 set to the timeout value 742 times out (step F103), the packet 721 and the packet 722 are aggregated. The created packet is created (step F104).
Here, the number of packets 422 to be aggregated into the aggregated packets is the same as the number aggregated in step F102.

  Next, the packet aggregating unit 73 transmits the created packet to the network 3 if the aggregated packet is created in step F102 or step F104 (step F105). Next, the packet aggregating unit 73 resets the timeout value of the timer 731 to a value determined by the timeout value 742 (step F106). Next, returning to the operation of step F101, the conventional packet transmission system continuously transmits the packet output by the packet output application 71 to the network 3.

  With such an operation, the conventional packet transmission system collects a plurality of packets and transmits them to the network to realize broadband packet transmission. As an example, when the network 3 is a network that conforms to a wireless LAN (Local Area Network) protocol, the time interval of packets that are continuously transmitted is determined by the network protocol. As a result, the bandwidth of packets that can be transmitted to the network 3 by the packet output application 71 is limited. However, by collecting and transmitting multiple packets, the corresponding time interval between the aggregated packets can be reduced, and the broadband Packet communication can be realized.

International Publication No. 2012/032606 JP-T-2006-526369 JP 2006-067566 A

K. Kim, S. Ganguly, R. Izmailov, S. Hong, "On Packet Aggregation Mechanisms for Improving VoIP Quality in Mesh Networks" IEEE 63rd Vehicular Technology Conference, May 7-10, 2006, pp. 891-895

  However, the non-patent document 1 has a problem that a transmission delay occurs in the packets when the packets are aggregated.

The reason for this is that when multiple packets are aggregated and sent to the network, the data length for continuous transmission to one destination increases, and the time interval for each destination packet to obtain a transmission opportunity increases. It is.
[Object of invention]
An object of the present invention is to provide a broadband packet transmission apparatus, a packet transmission system, and a packet transmission method while suppressing an increase in delay due to packet aggregation.

The packet transmission apparatus of the present invention comprises a storage means for classifying and storing packets for each destination;
Extracting means for extracting a plurality of packets for each destination stored in the storage means;
A maximum packet number setting means for setting the maximum number of packets that the extraction means collectively extracts,
The maximum number-of-packets setting means sets the maximum number of packets so that a total gain representing a bandwidth increase due to the extracted packet aggregation and a transmission delay due to the packet aggregation is a predetermined value or more. And

The packet transmission system of the present invention includes bridge means for collecting and transmitting packets, management means for managing the bridge means, and a network for connecting the bridge means and the management means,
The bridging means includes
Storage means for classifying and storing packets for each destination;
Extracting means for extracting a plurality of packets for each destination stored in the storage means;
Address storage means for holding an address used to encapsulate the packet extracted by the extraction means;
A packet encapsulation unit that retrieves an address to be used for encapsulation from the address storage unit and aggregates and encapsulates a plurality of packets extracted by the extraction unit;
The management means includes
Maximum packet number calculating means for calculating the maximum number of packets extracted by the extracting means at a time;
Parameter input means for inputting parameters used by the maximum packet number calculating means to calculate the maximum number of packets;
Transmitting means for transmitting the maximum number of packets calculated by the maximum packet number calculating means to the bridge means;
It is characterized by including.

According to the packet transmission method of the packet transmission apparatus of the present invention, packets are classified and stored for each destination,
Pull out a plurality of packets for each stored destination,
Performing the multiple extraction based on the maximum number of packets to be extracted together;
The maximum number of packets is determined such that a total gain representing a bandwidth increase due to the extracted packet aggregation and a transmission delay due to the packet aggregation is equal to or greater than a predetermined value.

  According to the present invention, an increase in delay due to packet aggregation can be suppressed, and at the same time, broadband packet transmission can be realized.

It is a block diagram which shows the structure of the packet transmission system of the 1st Embodiment of this invention. It is a block diagram which shows the structure of the packet transmission apparatus of the 1st Embodiment of this invention. It is a figure which shows three modes of a rate setting part. It is a figure for demonstrating the 1st Embodiment of this invention. It is a figure for demonstrating the 1st Embodiment of this invention. It is a figure for demonstrating the content of the encapsulation table. It is a flowchart which shows the operation | movement of the 1st Embodiment of this invention. It is a figure for demonstrating the 1st Embodiment of this invention. It is a flowchart which shows the operation | movement of the 1st Embodiment of this invention. It is a flowchart which shows the operation | movement of the 1st Embodiment of this invention. It is a flowchart which shows the operation | movement of the 1st Embodiment of this invention. It is a flowchart which shows the operation | movement of the 1st Embodiment of this invention. It is a block diagram which shows the structure of the packet transmission system of the 2nd Embodiment of this invention. It is a block diagram which shows the structure of the packet transmission apparatus of the 2nd Embodiment of this invention. It is a block diagram which shows the structure of a prior art. It is a flowchart which shows operation | movement of a prior art.

  The packet transmission apparatus of the present embodiment includes a buffer aggregation queue (21 in FIG. 2, corresponding to storage means), a packet extraction unit (22 in FIG. 2, corresponding to extraction means), and a maximum aggregation number calculation unit (FIG. 2 of 29, which corresponds to the maximum packet number setting means). The buffer aggregation queue 21 stores the packets in an internal queue for each destination. The packet extraction unit 22 searches the internal queue of the buffer aggregation queue by a round robin method, and extracts a plurality of packets of the same destination stored in the internal queue being searched. The maximum aggregation number calculation unit 29 performs calculation for optimizing the trade-off relationship between bandwidth increase and delay increase due to packet aggregation, and sets the maximum number of packets that the packet extraction unit extracts in one process. As a result, the packet transmission device extracts all the packets of the same destination stored in the buffer aggregation queue up to the maximum aggregation number determined by the maximum aggregation number calculation unit, and combines the extracted one or more packets into one packet. Operates to encapsulate and send to the network. The packet transmission apparatus of the present embodiment adopts such a configuration, determines the maximum number of packets to be aggregated in one process, and aggregates and transmits a plurality of packets.

  Next, a first embodiment for carrying out the present invention will be described in detail with reference to the drawings.

  Referring to FIG. 1, the packet transmission system according to the first embodiment of the present invention includes a packet output node 1-A, a packet output node 1-B, and a packet output node 1-C that perform transmission and reception of packets. Further, the packet transmission system according to the first embodiment of the present invention aggregates a plurality of packets output from the packet output node 1-A, the packet output node 1-B, and the packet output node 1-C, respectively, by the network 3 A packet aggregation device 2-A, a packet aggregation device 2-B, and a packet aggregation device 2-C. Furthermore, the packet transmission system according to the first embodiment of the present invention includes a network 3 that interconnects the packet aggregation device 2-A, the packet aggregation device 2-B, and the packet aggregation device 2-C. The packet output node 1-A to the packet output node 1-C communicate with each other via the network 3. The configurations of the packet output node 1-A to the packet output node 1-C are the same. The configurations of the packet aggregation device 2-A to the packet aggregation device 2-C are also the same. In this description, the case where there are three sets of packet output nodes and packet aggregation devices is shown. However, in this embodiment, this number is not limited. For example, the number of packet output nodes and packet aggregation devices is 1, There may be two or more than three.

  The packet aggregation device 2-A, the packet aggregation device 2-B, and the packet aggregation device 2-C are each a packet transmission device.

  FIG. 2 is a block diagram showing an internal configuration of the packet aggregation device 2-A that is a packet transmission apparatus according to an embodiment of the present invention. Since the configuration and operation of the packet aggregation device 2-B and the configuration and operation of the packet aggregation device 2-C are the same, only the packet aggregation device 2-A will be described below.

  Referring to FIG. 2, the packet aggregation device 2-A includes a packet aggregation queue 21, a packet extraction unit 22, a rate setting unit 23, a packet encapsulation unit 24, an encapsulation table 25, and a retransmission congestion control unit 26. A packet decapsulation unit 27, a maximum aggregation number calculation unit 29, and a trade-off parameter setting unit 28. The packet aggregation queue 21 stores the packets output by the packet output node 1-A for each destination in a plurality of internal queues. The packet extraction unit 22 extracts a plurality of packets of the same destination stored in the packet aggregation queue 21. The rate setting unit 23 sets a speed at which the packet extracting unit 22 extracts a packet. The packet encapsulation unit 24 encapsulates a plurality of packets extracted by the packet extraction unit 22 into a single packet. The encapsulation table 25 holds destination address information described in the header of the encapsulated packet when the packet encapsulating unit 24 encapsulates a plurality of packets. The retransmission congestion control unit 26 transmits a packet in which a plurality of packets are encapsulated to the network 3 at a speed at which congestion does not occur in the network 3. The packet decapsulation unit 27 receives a packet in which a plurality of packets are encapsulated from the retransmission congestion control unit 26, releases the encapsulation, returns it to the original plurality of packets, and transmits the packet to the packet output node 1-A . The maximum aggregation number calculation unit 29 sets the maximum number of packets that the packet extraction unit 22 extracts in one process. The trade-off parameter setting unit 28 gives parameters for the maximum aggregation number calculation unit 29 to calculate the maximum number of packets.

  The packet aggregation queue 21 is a storage unit that classifies and stores packets for each destination, the maximum aggregation number calculation unit 29 is a maximum packet number setting unit, the trade-off parameter setting unit is a parameter input unit, and the encapsulation table 25 is an address storage unit The packet encapsulation unit 24 corresponds to packet encapsulation means.

  The rate setting unit 23 operates the packet extracting unit 22 in any of the three modes shown in FIG. (A) In the physical link bandwidth mode 231, the rate setting unit 23 changes the packet collection unit 22 from the packet aggregation queue 21 to the round robin method at the speed of the bandwidth of the network link where the retransmission congestion control unit 26 and the network 3 are connected. Set to pull out the packet based on. At this time, the rate setting unit 23 notifies the packet extraction unit 22 of the bandwidth of the network link to which the retransmission congestion control unit 26 and the network 3 are connected. Here, the round robin method is also called “cyclical parallel processing”, and the process of extracting packets based on the round robin method is performed by sequentially searching a plurality of internal queues in the buffer aggregation queue to obtain a plurality of the same destinations. The process of pulling out the packet is executed.

  In (B) the congestion control bandwidth mode 232, the rate setting unit 23 performs setting so that packets are extracted from the packet aggregation queue 21 at a predetermined speed based on the round robin method. Note that the predetermined rate refers to the total packet transmission rate set by the packet extraction unit 22 for each destination by the retransmission congestion control unit 26 or a network in which the retransmission congestion control unit 26 and the network 3 are connected. The speed of the smaller of the bandwidth of the link. At this time, the rate setting unit 23 extracts the total packet transmission rate set for each destination by the retransmission congestion control unit 26 and the bandwidth of the network link where the retransmission congestion control unit 26 and the network 3 are connected. Notify 22 Further, in (C) individual flow band mode 233, the rate setting unit 23 transmits the packet of each destination at the same rate as the retransmission congestion control unit 26 transmits to the packet destination. Setting is made so that packets are extracted from the packet aggregation queue 21 based on the round robin method. At this time, the rate setting unit 23 notifies the packet extraction unit 22 of the rate at which the retransmission congestion control unit 26 is transmitting packets to each destination connected via the network 3.

  The packet aggregation queue 21 stores the packets output by the packet output node 1-A separately for each destination. For this reason, the packet aggregation queue 21 holds as many internal queues as the number of destination nodes. The packet aggregation queue 21 classifies the packets output by the packet output node 1-A for each destination in the order of reception, and stores them in a queue divided for each destination node.

  The packet extracting unit 22 operates in any one of the three modes set by the rate setting unit 23. In (A) physical link bandwidth mode 231 and (B) congestion control bandwidth mode 232, the packet extraction unit 22 changes from the packet aggregation queue 21 to the round robin method at the speed (data capacity / unit time) specified by the rate setting unit 23. Pull out the packet based on. As shown in FIG. 4, in packet extraction based on round robin, the packet extraction unit 22 searches the internal queue for each destination held in the packet aggregation queue 21 in order. At this time, if the same destination packet is stored in the searched internal queue, the packet extraction unit 22 extracts all the stored packets to the extraction packet encapsulation unit 24. However, the upper limit value of the number of packets to be extracted from the internal queue of the packet aggregation queue 21 is the value set by the maximum aggregation number calculation unit 29 and the total size of the packets to be extracted exceeds the upper limit value of the packet size determined by the network 3 There is no maximum number and the smaller one. Here, the upper limit value of the number of packets indicated by the maximum aggregation number calculation unit 29 can be indicated by the number of packets or the total size of the aggregated packets.

  In addition, in (C) individual flow bandwidth mode 233, the packet extraction unit 22 extracts each destination packet from the packet aggregation queue 21 at the transmission rate to the network 3 set for each destination by the retransmission congestion control unit 26. . As shown in FIG. 5, the packet extraction unit 22 internally holds a per-flow credit buffer 221 and a credit recalculation timer 222 for operation in (C) individual flow band mode 233. The credit buffer 221 for each flow holds an element for each destination with which the packet output node 1-A communicates. In the credit buffer 221 for each flow, credits corresponding to the transmission speed for each destination notified by the packet extracting unit 22 from the rate setting unit 23 are accumulated over time. Here, credit is a right to transmit a unit data length.

  The credit recalculation timer 222 is a timer for managing the period in which credits are accumulated in the credit buffer 221 for each flow. The packet extraction unit 22 searches the internal queue of the packet aggregation queue 21 by round-robin method, and refers to the credit buffer 221 for each flow corresponding to the destination corresponding to the internal queue, and stores it in the credit buffer 221. If the received credit is equal to or larger than the total size of the packets stored in the internal queue, all the packets stored in the internal queue are extracted and passed to the packet encapsulation unit 24. Further, the packet extracting unit 22 subtracts the credit value corresponding to the total packet size extracted from the internal queue of the packet aggregation queue 21 from the element corresponding to the packet destination of the per-flow credit buffer 221. However, if the number of stored packets exceeds the smaller value of the value set by the maximum aggregation calculation unit 29 or the upper limit value of the packet size determined by the network 3, the smaller value will be This is the upper limit of the number of packets to be extracted during processing. Therefore, if the credit stored in the credit buffer 221 is equal to or greater than the upper limit value of the number of packets, the upper limit value number of packets stored in the internal queue are extracted. Further, the packet extracting unit 22 subtracts the credit value corresponding to the total size of the upper limit packet extracted from the internal queue of the packet aggregation queue 21 from the element corresponding to the packet destination of the credit buffer 221 for each flow.

  The encapsulation table 25 holds address information described in the header of a packet to be encapsulated when the packet transmitted by the packet output node 1-A is encapsulated. As shown in FIG. 6, the encapsulation table 25 associates the destination address of the packet to be encapsulated with the destination address of the packet transmitted by the packet output node 1-A. At this time, the packet destination address may be a single value, or in a certain address area so that the destination memory address used as the destination address in the packet of the PCI (Peripheral Component Interconnect) Express bus is an example. In this case, the packet destination address area may be used as an element of the packet destination address in the encapsulation table 25, and the encapsulation address may be associated with the area.

  The packet encapsulation unit 24 receives one or more packets of the same destination extracted together by the packet extraction unit 22, refers to the encapsulation table 25, and uses the destination of the packet received from the packet extraction unit 22 as a search key. Retrieve the address used for encapsulation. The packet encapsulation unit 24 encapsulates one or more packets received from the packet extraction unit 22 using the retrieved address for encapsulation, and encapsulates the packet as a single packet, which is passed to the retransmission congestion control unit 26.

  The retransmission congestion control unit 26 transmits the packet received from the packet encapsulation unit 24 to the network 3 while adjusting the transmission speed so as not to cause congestion in the network 3. Therefore, the retransmission congestion control unit 26 performs control to transmit a packet for each destination connected to the network 3. The retransmission congestion control unit 26 is a rate setting unit 23 that transmits the packet transmission rate for each destination, the total value of those packet transmission rates, and all or some of the physical bandwidth of the network link connected to the network 3. Notify In addition, when the packet transmitted to the network 3 does not reach the destination due to congestion in the network 3 or a bit error, the retransmission congestion control unit 26 retransmits the packet. The retransmission congestion control unit 26 also passes a packet in which a plurality of packets received from the network 3 are encapsulated to the packet decapsulation unit 27.

The trade-off parameter setting unit 28 sets parameters used to calculate the optimal number of aggregations according to the trade-off relationship between the increase in bandwidth by consolidating and encapsulating packets and the increase in delay. (Hereafter, this parameter is referred to as γ).
The parameter γ is a real number indicating the relative importance of gain due to bandwidth improvement and loss due to increased delay, and is a value between negative infinity and positive infinity. A system user or administrator uses the trade-off parameter setting unit 28 to set the parameter γ.

  The maximum aggregation number calculation unit 29 calculates the optimum value of the maximum number of packets to be aggregated and encapsulated into one packet using the parameter γ given from the trade-off parameter setting unit 28, and the value is calculated by the packet extraction unit 22 Set to. Here, the maximum number of packets may be set by using the number of packets, or the total size of the aggregated packets may be used. The maximum aggregation number calculation unit 29 obtains an optimum value by obtaining n that maximizes the calculation result of the following optimization expression 1.

  In the optimization formula 1, n is the maximum number of packets to be aggregated into one packet, O is the overhead when the packets are individually encapsulated, and γ is a parameter given from the trade-off parameter setting unit 28. Here, the overhead refers to the data length added to the packet for encapsulation.

The first term of the optimization formula 1 indicates a bandwidth improvement rate by consolidating and encapsulating packets. The second term indicates the rate of increase in delay. The increase rate of the delay is proportional to the number of packets to be aggregated.
This is because the packet extraction unit 22 extracts packets for each destination from the packet aggregation queue 21 in a round-robin manner, so that each destination packet is given a transmission opportunity in proportion to the number of packets extracted from each destination queue. This is because the time interval until is increased.
The optimization equation 1 is obtained by calculating a linear sum of both elements with a bandwidth increase rate as a gain element, a delay increase rate as a loss element, and a parameter γ as a coefficient that explains the balance of contributions of both elements. Give total gain. Therefore, by obtaining n that maximizes the optimization expression 1 indicating the gain, the maximum number of packets n that the packet extraction unit 22 extracts in a single process is obtained. Here, the maximum aggregation number calculation unit 29 may actually calculate using the given parameter γ to obtain n that maximizes the calculation result of the optimization formula 1, but in addition to the value of O and γ It is also possible to use a method in which a table indicating n for maximizing the calculation result of the optimization formula 1 is stored in correspondence with the value of and the optimal value n is obtained by referring to the table. Note that here, the number of packets that the packet extraction unit 22 extracts together in one process is n, which maximizes the calculation result of the optimization formula 1 (overall gain). The maximum number of packets to be extracted may not be a value that maximizes the calculation result of the optimization formula 1 (total gain), but is a number that is smaller than n and has a total gain that is equal to or greater than a predetermined value. May be set. This is because the overall gain may not be maximized depending on the application.

  Here, the total gain calculates the sum of both elements when γ is positive, but calculates the difference between both elements when γ is negative.

  Instead of the optimization formula 1, the maximum number of packets may be obtained using the following optimization formula 2. N is the number of destinations with which the packet aggregation device A 2 -A is communicating. The reason for multiplying N in the second term is that the delay with which the destination packet of the round robin method can obtain a transmission opportunity increases in proportion to the number of destinations.

  Next, the operation of the embodiment of the present invention will be described with reference to the drawings. First, the operation in which the packet output node 1-A transmits a packet to the network 3 in the embodiment of the present invention will be described with reference to FIGS. FIG. 7 is a flowchart showing an operation in which the packet output node 1-A transmits a packet to the network 3.

First, the packet output node 1-A outputs a packet (step A101). Next, the packet aggregation queue 21 receives the packet from the packet output node 1-A,
The received packet is stored in the internal queue corresponding to the packet output node that is the destination of the received packet (step A102). Subsequently, the packet extraction unit 22 operates in any one of the (A) physical link bandwidth mode 231, (B) congestion control bandwidth mode 232, and (C) individual flow bandwidth mode 233. Here, in (A) physical link bandwidth mode 231 and (B) congestion control bandwidth mode 232, the packet extraction unit 22 searches the internal queue of the packet aggregation queue 21 in a round based on the round robin method. The packet destined for the same destination is extracted with a predetermined maximum value as an upper limit (step A103). The maximum value determined here is the value set by the maximum aggregation number calculation unit 29, and the total size of the packets stored in the internal queue of the same destination is the upper limit value of the packet size determined by the network 3 The smaller of the maximum values not exceeding. Also, in the operation of (C) individual flow bandwidth mode 233, the packet extraction unit 22 searches the internal queue of the packet aggregation queue 21 in a round based on the round robin method. Is compared with the total size of the packets destined for the same destination and the remaining credit described in the element corresponding to the destination of the credit buffer 221 for each flow. All packets are withdrawn from the internal buffer in search of the aggregation queue 21. The credit corresponding to the extracted packet is subtracted from the element corresponding to the destination of the credit buffer 221 for each flow. However, if the total size of the packets stored in the internal queue of the same destination exceeds the specified maximum value described in (A) Physical link bandwidth mode 233 and (B) Congestion control bandwidth mode 232, The total size of the maximum number of packets is compared with the elements of the credit buffer 221 for each flow, and the presence / absence of packet extraction is determined. When it is determined that the packet is to be extracted, the maximum number of packets is extracted, and the credit corresponding to the extracted packet is subtracted from the element corresponding to the destination credit buffer 221. Next, the packet encapsulating unit 24 refers to the encapsulation table 24 and searches for an encapsulated address using the destination addresses of one or more packets received from the packet extracting unit 22 as search keys (step A104). Subsequently, the packet encapsulation unit 24 encapsulates one or more packets received from the packet extraction unit 22 into one packet using the retrieved address as shown in FIG. 8 (step A105). The encapsulation header is used to encapsulate a plurality of packets, and the address searched in step A104 is used as the destination address of the encapsulation header. Subsequently, the retransmission congestion control unit 26 receives the packet from the packet encapsulation unit 24, and transmits the packet to the network 3 at a transmission rate determined for each destination packet output node (step A106).

  Next, the operation of the packet extracting unit 22 in (A) the physical link bandwidth mode 231 and (B) the congestion control bandwidth mode 232 will be described with reference to FIGS. FIG. 9 is a flowchart showing the operation of the packet extraction unit 22 in (A) physical link bandwidth mode 231 and (B) congestion control bandwidth mode 232.

  First, the rate setting unit 23 determines the round robin operation speed (data to be extracted / unit time) of the packet extracting unit 22 (step B101). The round robin speed is the bandwidth of the network link that the retransmission congestion control unit 26 connects to the network 3 in the (A) physical link bandwidth mode 231, and the retransmission congestion control unit 26 in the (B) congestion control bandwidth mode 232. It is the smaller value of the total value of the transmission rates set for each destination and the bandwidth of the network link where the retransmission congestion control unit 26 is connected to the network 3. Next, the packet extracting unit 22 refers to the first internal queue of the packet aggregation queue 21, extracts all the packets stored in the queue, and passes them to the packet encapsulation unit (step B102). However, when the number of packets stored in the internal queue exceeds the maximum aggregation number, the maximum aggregation number of packets is extracted from the internal queue. Here, the maximum aggregation number is the smaller of the value set in the maximum aggregation number calculation unit 29 and the maximum value where the total size of the aggregated packets does not exceed the upper limit value of the packet size determined by the network 3 Is the value of Next, the packet extracting unit 22 checks whether or not the internal queue from which the packet has been extracted is the last internal queue held by the packet aggregation queue 21 (step B103). The packet is extracted from the queue (step B104). The operation of step B104 is the same as the operation of extracting the packet in step B102. Subsequently, the packet extraction unit 22 returns to step B103, and repeats steps B103 and B104 to repeat extraction of packets up to the last internal queue held by the packet aggregation queue 21. When the packet withdrawal reaches the last internal queue, the conditional branch at step B103 determines Yes and returns to step B102, whereby the packet withdrawal order returns to the first internal queue of the packet aggregation queue 21. Thereafter, the operations after Step B102 are repeated. In (B) the congestion control bandwidth mode 232, it is possible to update the round robin operation speed of the packet extraction unit 22 determined in step B101 at regular intervals. In this case, for the calculation of the update value of the rate, the transmission rate determined by the retransmission congestion control unit 26 at that time for each destination is used.

  Next, the operation of the packet extracting unit 22 in (C) the individual flow band mode 233 will be described with reference to FIGS. 5, 10, and 11. FIG. 10 is a flowchart showing an operation of extracting a packet from the packet aggregation queue 21 of the packet extracting unit 22 in the (C) individual flow band mode 233. FIG. 11 is a flowchart showing an operation for updating credits in the credit buffer 221 for each flow in the packet extracting unit 22.

  First, the packet extracting unit 22 confirms the value of the remaining credit of the destination corresponding to the first internal queue of the packet aggregation queue 21 with reference to the element corresponding to the destination of the credit buffer 221 for each flow (step C101). . Subsequently, if the value of the confirmed credit is equal to or larger than the total size of the packets stored in the first internal queue of the packet aggregation queue 21 (step C102), the packet extraction unit 22 extracts all packets from the internal queue (step S102). C103). Then, the credit value corresponding to the total size of the extracted packets is subtracted from the element corresponding to the destination of the credit buffer 221 for each flow (step C104). Here, the packet extracted by the packet extracting unit 22 from the internal buffer of the packet aggregation queue 21 may be a predetermined maximum number of packets instead of all the packets accumulated in the internal queue. This is the same as the operation described in step A103. In this case, the value that the packet extraction unit 22 compares with the credit value in step C102 is the total size of the maximum number of packets, and the credit value to be subtracted from the element corresponding to the destination of the credit buffer 221 for each flow is the maximum number of packets. The credit value corresponding to the total size. Subsequently, the packet extracting unit 22 confirms whether the internal queue from which the packet has been extracted is the last internal queue held by the packet aggregation queue 21 (step C105), and if it is not the last queue, the packet is extracted to the next internal queue destination. The element of the corresponding credit buffer 221 for each flow is confirmed (step C106). Next, if the value of the confirmed credit is equal to or larger than the total size of the packets stored in the internal queue of the destination corresponding to the confirmed credit in the packet aggregation queue 21 (step C107), the packet extracting unit 22 Is extracted (step C108). Then, the credit value corresponding to the total size of the extracted packets is subtracted from the element corresponding to the destination of the credit buffer 221 for each flow (step C109). Here, if the number of packets stored in the aggregation queue 21 exceeds the predetermined maximum number of packets, the number of packets extracted by the packet extraction unit 22 is the maximum number of packets, which is the same as the operation in step C103, and the credit to be subtracted is the maximum. The value corresponding to the number of packets is the same as the operation in step C104. Subsequently, the packet extracting unit 22 repeats the operations of Step C105 to Step C109 until reaching the last internal queue held by the packet aggregation queue 21. Then, after reaching the last internal queue, the packet extraction unit 22 returns to the first internal queue held by the packet aggregation queue 21 (step C105), and thereafter repeats the operations from step C101 to step C109.

  In addition, the packet extracting unit 22 periodically updates the credit buffer 221 for each flow in parallel with the operation of transmitting the packet received from the packet output node 1-A to the network 3 by the operations of Step C101 to Step C109. Do. The packet extraction unit 22 receives the transmission rate set for each destination node by the retransmission congestion control unit 26 from the rate setting unit 23, and uses a credit value depending on the transmission rate of each destination to provide a credit buffer for each flow. 221 is updated (step D101). Next, the packet extraction unit 22 sets the credit recalculation timer 222 to a predetermined value (step D102). Subsequently, the packet extracting unit 22 waits for a timeout of the credit recalculation timer 222 (step D103), and returns to step D101 after the timeout. In the subsequent operation of step D101, the retransmission congestion control unit 26 receives a new transmission rate set for each destination from the rate setting unit 23 and uses the value for credit calculation.

  Next, the operation for setting the maximum number of packets collected by the packet extraction unit 22 will be described with reference to FIGS. FIG. 12 is a flowchart showing an operation in which the maximum aggregation number calculation unit 29 sets the maximum number of packets aggregated by the packet extraction unit 22.

  The trade-off parameter setting unit 28 receives an input of the parameter γ used for calculating the optimum maximum aggregate packet number according to the trade-off relationship between the increase in bandwidth and the increase in delay due to packet aggregation (step E101). The input parameter γ is set by a system user or administrator. The maximum aggregation number calculation unit 29 uses the parameter γ received from the trade-off parameter setting unit 28 to obtain n (n is an integer) that maximizes the optimization formula 1 as the maximum aggregation number (step E102). Subsequently, the maximum aggregation number calculation unit 29 sets the obtained maximum packet aggregation number in the packet extraction unit 22 (step E103).

  In the embodiment of the present invention, the packet extracting unit extracts and encapsulates all the packets with the maximum number of packets set by the maximum aggregation number calculating unit as an upper limit, and transmits the packet to the network. By adopting such a configuration, a plurality of packets are aggregated and encapsulated and transmitted to the network, so that the network resulting from the encapsulation header is compared with the case where the packets are individually encapsulated and transmitted to the network. Bandwidth overhead can be reduced, and packets can be transmitted over a wide band. At the same time, by setting the maximum number of packets to be aggregated, an increase in packet transmission delay due to packet aggregation can be prevented, and packets can be transmitted with low delay.

  Next, a second embodiment for carrying out the present invention will be described with reference to FIGS. FIG. 13 is a diagram showing the configuration of the packet transmission system of this embodiment, and FIG. 14 is a diagram showing the configuration of the packet aggregation device 6-A.

  Referring to FIGS. 13 and 14, the packet aggregation device 6-A replaces the trade-off parameter setting unit 28 and the maximum aggregation number calculation unit 29 held by the packet aggregation device 2-A shown in FIG. A management manager communication unit 61 for communication is included. The management manager communication unit 61 sets the maximum packet aggregation number received from the management manager 5 in the packet extraction unit 22 as the maximum number of packets that the packet extraction unit 22 extracts from the packet aggregation queue in a single process. The packet aggregation device 6-A serves as a bridge means, and the management manager 5 corresponds to a management means.

  Further, the configuration of the packet aggregation device 6-B and the configuration of the packet aggregation device 6-C are the same as the configuration of the aggregation device 6-A, and constitute a bridging means. Further, the combination of the packet output node and the packet aggregation device is not limited to three, which is the same as in the first embodiment.

  The management manager 5 has a trade-off parameter setting interface 51 for receiving the setting of the parameter γ used to calculate the optimum maximum packet aggregation number from the trade-off between communication bandwidth increase and delay increase due to packet aggregation from the system user and administrator; Using the set parameter γ, the maximum aggregation number calculation program 52 for obtaining n (n is an integer) that maximizes the optimization expression 1 as the maximum aggregation number, and the calculated maximum packet aggregation number using the network 3 A communication program 53 for notifying all packet aggregation devices. The communication program 53 corresponds to transmission means.

  By adopting such a configuration, the optimum maximum packet aggregation number is collectively calculated by the management manager, transmitted to each packet aggregation device using the network, and the maximum packet aggregation number of the packet aggregation device is set. This makes it easy to unify and manage the maximum number of packets aggregated in the system, and at the same time, performs calculation of optimization formula 1 that includes division and is complicated to implement in hardware by the management manager software By doing so, the circuit of the packet aggregation device can be simplified.

  Although the embodiments of the present invention have been described above, the packet extraction unit 22, the rate setting unit 23, the encapsulation unit 24, the packet decapsulation unit 27, and the packet aggregation device shown in FIGS. The trade-off parameter unit 28 and the maximum aggregation number calculation unit 29 may be configured by hardware or software. Some or all of the functions of each part are read by a computer from a computer-readable recording medium such as a CD-ROM (Compact Disc Read Only Memory), DVD (Digital Versatile Disc), or flash memory. It can also be realized by executing.

  For example, a computer as a packet aggregation device is used to store a program, a storage unit such as a hard disk, a ROM (Read Only Memory), a display unit such as a liquid crystal display, a DRAM (Dynamic Random Access Memory) that stores data necessary for calculation. ), A CPU (Central Processing Unit), and a bus that connects each unit. The operation of each unit shown in FIGS. 2, 5, and 14 is described by a program, this program is stored in a storage unit such as a ROM, information necessary for calculation is stored in a DRAM, and the program is operated by a CPU. Thus, the function of the packet aggregation device according to the present embodiment can be realized by a program.

  Furthermore, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the gist of the present invention described above.

  A part or all of the above embodiment can be described as in the following supplementary notes, but is not limited to the following configuration.

(Appendix 1)
Storage means for classifying and storing packets for each destination;
Extracting means for extracting a plurality of packets for each destination stored in the storage means;
A maximum packet number setting means for setting the maximum number of packets that the extraction means collectively extracts,
The maximum number-of-packets setting means sets the maximum number of packets so that a total gain representing a bandwidth increase due to the extracted packet aggregation and a transmission delay due to the packet aggregation is a predetermined value or more. A packet transmission device.

(Appendix 2)
A parameter input means for inputting a parameter used by the maximum packet number setting means to calculate the maximum number of packets to the maximum packet number setting means;
The maximum packet number setting means includes a first term that gives the parameter input from the parameter input means, including the maximum number of packets and an increase in bandwidth due to packet aggregation, and the maximum packet number, and a transmission delay due to packet aggregation. The sum or difference between the first term and the second term is used as a coefficient that gives a relative magnitude with the second term that gives an increase, and the total gain is maximized. The packet transmission apparatus according to appendix 1, wherein the maximum number of packets is determined as described above, and the determined maximum number of packets is set as the maximum number of packets.

(Appendix 3)
The packet transmission apparatus according to appendix 2, wherein the second term is proportional to the maximum number of packets.

(Appendix 4)
An address storage means for holding an address used for collecting and encapsulating the packets extracted together by the extraction means;
A packet encapsulation unit that retrieves an address to be used for encapsulation from the address storage unit and aggregates and encapsulates a plurality of packets extracted by the extraction unit;
The packet transmission device according to any one of appendices 1 to 3, further comprising:

(Appendix 5)
Bridge means for collecting and transmitting packets, management means for managing the bridge means, and a network for connecting the bridge means and the management means,
The bridging means includes
Storage means for classifying and storing packets for each destination;
Extracting means for extracting a plurality of packets for each destination stored in the storage means;
Address storage means for holding an address used to encapsulate the packet extracted by the extraction means;
A packet encapsulation unit that retrieves an address to be used for encapsulation from the address storage unit and aggregates and encapsulates a plurality of packets extracted by the extraction unit;
The management means includes
Maximum packet number calculating means for calculating the maximum number of packets extracted by the extracting means at a time;
Parameter input means for inputting parameters used by the maximum packet number calculating means to calculate the maximum number of packets;
Transmitting means for transmitting the maximum number of packets calculated by the maximum packet number calculating means to the bridge means;
A packet transmission system comprising:

(Appendix 6)
Packets are classified and stored for each destination,
Pull out a plurality of packets for each stored destination,
Performing the multiple extraction based on the maximum number of packets to be extracted together;
The maximum number of packets is determined such that a total gain representing a bandwidth increase due to extracted packet aggregation and a transmission delay due to the packet aggregation is equal to or greater than a predetermined value,
A packet transmission method for a packet transmission apparatus.

(Appendix 7)
Calculating the maximum number of packets with the parameter giving the maximum number of packets to be pulled together,
The parameter is used as a coefficient that gives a relative magnitude between a first term that includes the maximum number of packets and that increases bandwidth due to packet aggregation and a second term that includes the maximum number of packets and increases transmission delay due to packet aggregation. Then, the sum or difference between the first term and the second term is defined as a total gain, the maximum number of packets is determined so that the total gain is maximized, and the determined maximum number of packets is determined as the maximum number of packets. The packet transmission method according to appendix 6, characterized in that the maximum number is set.

(Appendix 8)
The packet transmission method according to appendix 7, wherein the second term is proportional to the maximum number of packets.

(Appendix 9)
Holding an address used for encapsulating the extracted packet in an address storage means;
The packet transmission method according to any one of appendices 6 to 8, wherein an address used for encapsulation is searched from the address storage unit, and the plurality of extracted packets are aggregated and encapsulated.

(Appendix 10)
A packet transmission method for a packet transmission apparatus, comprising: bridge means for collecting and transmitting packets; management means for managing the bridge means; and a network connecting the bridge means and the management means,
The bridging means includes
Packets are classified and stored for each destination,
Pull out a plurality of packets for each stored destination,
Holding an address used for encapsulating the pulled packet in an address storage means;
An address used for encapsulation is retrieved from the address storage means, and the plurality of extracted packets are aggregated and encapsulated,
Furthermore, the management means includes
Calculating the maximum number of packets that the bridging means pulls at a time;
Accept parameters used to calculate the maximum number of packets,
A packet transmission method, wherein the calculated maximum number of packets is transmitted to the bridge means using a network.

(Appendix 11)
To a computer as a packet transmission device,
Processing to classify and store packets for each destination;
A process of pulling out a plurality of packets stored for each destination together;
And a process for setting the maximum number of packets to be pulled out together,
A packet transmission program in which the maximum number of packets is determined such that a total gain representing a bandwidth increase due to extracted packet aggregation and a transmission delay due to the packet aggregation is equal to or greater than a predetermined value.

(Appendix 12)
Calculating the maximum number of packets with the parameter giving the maximum number of packets to be pulled together,
The parameter is used as a coefficient that gives a relative magnitude between a first term that includes the maximum number of packets and that increases bandwidth due to packet aggregation and a second term that includes the maximum number of packets and increases transmission delay due to packet aggregation. Then, the sum or difference between the first term and the second term is defined as a total gain, the maximum number of packets is determined so that the total gain is maximized, and the determined maximum number of packets is The program for packet transmission according to appendix 11, wherein the maximum number is set.

(Appendix 13)
13. The packet transmission program according to appendix 12, wherein the second term is proportional to the maximum number of packets.

(Appendix 14)
Holding an address used for encapsulating the extracted packet in an address storage means;
The program for packet transmission according to any one of appendices 11 to 13, wherein an address used for encapsulation is searched from the address storage means, and the plurality of extracted packets are aggregated and encapsulated.

  While the present invention has been described with reference to the embodiments, the present invention is not limited to the above embodiments. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

  This application claims the priority on the basis of Japanese application Japanese Patent Application No. 2014-097918 for which it applied on May 9, 2014, and takes in those the indications of all here.

  INDUSTRIAL APPLICABILITY According to the present invention, the present invention can be applied to a use in which packets are transmitted with wide bandwidth and low delay between nodes communicating with each other through a network. Further, in a bus system in which a computer and an I / O device are connected using a network, the present invention can be applied to an application such as transmitting an I / O packet with a wide bandwidth and low delay.

1-A to 1-C packet output node
2-A to 2-C packet aggregation device
3 network
5 Manager
6-A to 6-C packet aggregation device
7-A to 7-C network node
21 Packet aggregation queue
22 Packet extraction part
23 Rate setting section
24 Packet encapsulation unit
25 Encapsulation table
26 Retransmission congestion controller
27 Packet decapsulation part
28 Trade-off parameter setting section
29 Maximum aggregate number calculation part
51 Tradeoff parameter setting interface
52 Maximum aggregation number calculation program
53 Communication program
61 Management Manager Communication Department
71 Packet output application
72 packet aggregation queue
73 Packet aggregator
74 memory
75 Packet receiver

Claims (10)

Storage means for classifying and storing packets for each destination;
Extracting means for extracting a plurality of packets for each destination stored in the storage means;
A maximum packet number setting means for setting the maximum number of packets that the extraction means collectively extracts,
The maximum number-of-packets setting means sets the maximum number of packets so that a total gain representing a bandwidth increase due to the extracted packet aggregation and a transmission delay due to the packet aggregation is a predetermined value or more. A packet transmission device. A parameter input means for inputting a parameter used by the maximum packet number setting means to calculate the maximum number of packets to the maximum packet number setting means;
The maximum packet number setting means includes a first term that gives the parameter input from the parameter input means, including the maximum number of packets and an increase in bandwidth due to packet aggregation, and the maximum packet number, and a transmission delay due to packet aggregation. The sum or difference between the first term and the second term is used as a coefficient that gives a relative magnitude with the second term that gives an increase, and the total gain is maximized. 2. The packet transmission apparatus according to claim 1, wherein the maximum number of packets is determined as described above, and the determined maximum number of packets is set as the maximum number of packets.   The packet transmission apparatus according to claim 2, wherein the second term is proportional to the maximum number of packets. An address storage means for holding an address used for collecting and encapsulating the packets extracted together by the extraction means;
A packet encapsulation unit that retrieves an address to be used for encapsulation from the address storage unit and aggregates and encapsulates a plurality of packets extracted by the extraction unit;
The packet transmission device according to claim 1, further comprising: Bridge means for collecting and transmitting packets, management means for managing the bridge means, and a network for connecting the bridge means and the management means,
The bridging means
Storage means for classifying and storing packets for each destination;
Extracting means for extracting a plurality of packets for each destination stored in the storage means;
Address storage means for holding an address used to encapsulate the packet extracted by the extraction means;
A packet encapsulation unit that retrieves an address to be used for encapsulation from the address storage unit and aggregates and encapsulates a plurality of packets extracted by the extraction unit;
The management means includes
Maximum packet number calculating means for calculating the maximum number of packets extracted by the extracting means at a time;
Parameter input means for inputting parameters used by the maximum packet number calculating means to calculate the maximum number of packets;
Transmitting means for transmitting the maximum number of packets calculated by the maximum packet number calculating means to the bridge means;
A packet transmission system comprising: Packets are classified and stored for each destination,
Pull out a plurality of packets for each stored destination,
Performing the multiple extraction based on the maximum number of packets to be extracted together;
The maximum number of packets is determined such that a total gain representing a bandwidth increase due to extracted packet aggregation and a transmission delay due to the packet aggregation is equal to or greater than a predetermined value,
A packet transmission method for a packet transmission apparatus. Calculating the maximum number of packets with the parameter giving the maximum number of packets to be pulled together,
The parameter is used as a coefficient that gives a relative magnitude between a first term that includes the maximum number of packets and that increases bandwidth due to packet aggregation and a second term that includes the maximum number of packets and increases transmission delay due to packet aggregation. Then, the sum or difference between the first term and the second term is defined as a total gain, the maximum number of packets is determined so that the total gain is maximized, and the determined maximum number of packets is determined as the maximum number of packets. The packet transmission method according to claim 6, wherein the maximum number of packets is set.   The packet transmission method according to claim 7, wherein the second term is proportional to the maximum number of packets. Holding an address used for encapsulating the extracted packet in an address storage means;
The packet transmission method according to any one of claims 6 to 8, wherein an address used for encapsulation is searched from the address storage unit, and the plurality of extracted packets are aggregated and encapsulated. A packet transmission method for a packet transmission apparatus, comprising: bridge means for collecting and transmitting packets; management means for managing the bridge means; and a network connecting the bridge means and the management means,
The bridging means includes
Packets are classified and stored for each destination,
Pull out a plurality of packets for each stored destination,
Holding an address used for encapsulating the pulled packet in an address storage means;
An address used for encapsulation is retrieved from the address storage means, and the plurality of extracted packets are aggregated and encapsulated,
Furthermore, the management means includes
Calculating the maximum number of packets that the bridging means pulls at a time;
Accept parameters used to calculate the maximum number of packets,
A packet transmission method, wherein the calculated maximum number of packets is transmitted to the bridge means using a network.

Images